Finite-size secret-key rates of discrete modulation continuous-variable quantum key distribution under Gaussian attacks

This paper derives analytical or semi-analytical bounds on finite-size secret-key rates for discrete-modulation continuous-variable quantum key distribution under Gaussian attacks by computing Petz-Rényi and sandwiched Rényi conditional entropies, offering tighter estimates for short block sizes compared to existing methods.

The Gist

Imagine you and a friend are trying to send secret messages to each other across a noisy, crowded room. You want to be sure that no one else (let's call her "Eve") can listen in and steal your secrets. This is the basic idea of Quantum Key Distribution (QKD): using the laws of physics to create a secret code that is impossible to crack without being detected.

This paper is like a safety manual for a specific type of high-tech lock called "Continuous-Variable QKD with Discrete Modulation." Here is a breakdown of what the authors did, using simple analogies.

1. The Setup: Sending Messages with Light Bulbs

In this system, Alice (the sender) sends messages using laser beams (coherent states).

• Discrete Modulation: Instead of sending a smooth, continuous stream of light, Alice turns the laser on and off in specific, distinct patterns. Think of it like Morse code, but using the phase (the timing) of the light wave.
  • BPSK (Binary): Like a light switch that is either "On" or "Off" (or shifted 180 degrees).
  • QPSK (Quadrature): Like a compass with four directions (North, East, South, West).
• Bob (the receiver): He catches the light and measures it to decode the message.

2. The Problem: The "Finite Size" Dilemma

In the real world, you can't send infinite messages. You send a "block" of messages, say 1,000 or 1,000,000, and then stop to check if the key is safe.

• The Old Way: Most safety checks assume you send an infinite number of messages. It's like saying, "If you flip a coin a billion times, you'll get 50% heads."
• The Reality: In the real world, you only flip the coin a few hundred times. If you get 60% heads, is the coin rigged, or was it just bad luck?
• The Challenge: When the block of messages is small (short distance or low speed), the "bad luck" (statistical noise) makes it very hard to prove the key is safe. The old math says, "We can't guarantee safety with such a small block," so the key rate drops to zero.

3. The Solution: New Mathematical "Rulers"

The authors developed new mathematical tools to measure safety more accurately when the block size is small. They used something called Rényi Entropies.

• The Analogy: Imagine you are trying to guess the weight of a bag of apples.
  • The Old Ruler (Von Neumann Entropy/AEP): This is a standard, heavy ruler. It works great if you have a huge pile of apples (infinite data). But if you only have 5 apples, the ruler is too clumsy; it gives you a huge margin of error, so you can't be sure of the weight.
  • The New Ruler (Sandwiched Rényi Entropy): This is a laser-precise, flexible measuring tape. It can wrap around a small pile of 5 apples and give you a much tighter, more accurate estimate.

The paper shows that for very small blocks of data (short distances), this new "laser tape" (Sandwiched Rényi Entropy) gives a much better estimate of how secure the key is. It proves that you can still generate a secret key even when the block size is too small for the old methods to work.

4. The Scenarios: The "Clean" Room vs. The "Noisy" Room

The authors tested their new ruler in two different environments:

• Scenario A: The Pure Loss Channel (The Long Hallway)
  Imagine the hallway is perfectly quiet, but very long. The light gets dimmer the further it travels (attenuation). Eve is just standing at the end of the hallway, catching the light that leaks out.

  • Result: The new ruler works beautifully here. It shows that even with significant signal loss, you can still get a secret key, especially with the new math.

• Scenario B: The Thermal Noisy Channel (The Stormy Room)
  Now imagine the hallway is windy and raining. There is extra "noise" (thermal photons) messing up the signal. Eve isn't just catching leaked light; she is actively throwing noise into the system to confuse Bob.

  • Result: This is harder. The math gets messy because the "bag of apples" is now wet and squishy. The authors had to use a "truncation" trick (ignoring the extremely rare, huge apples) to make the math work. Even with this noise, their new method showed better results than previous methods, allowing for secure keys over longer distances than before.

5. The Big Takeaway

The main point of this paper is: "Don't throw away your secret keys just because you don't have enough data yet."

Previous methods were too conservative. They would say, "Your block size is too small; you can't have a secret key." The authors say, "Not necessarily! If you use our new, sharper mathematical tools (Sandwiched Rényi Entropy), you can squeeze a secure key out of much smaller blocks of data."

In summary:

• Old Math: "You need a billion messages to be safe."
• New Math: "You only need a few thousand, and we can prove it's safe."

This is a huge step forward for making quantum encryption practical for real-world, short-distance applications (like securing a connection between two buildings in a city) where you can't wait to send billions of messages.

Technical Summary

Here is a detailed technical summary of the paper "Finite-size secret-key rates of discrete modulation continuous-variable quantum key distribution under Gaussian attacks."

1. Problem Statement

Quantum Key Distribution (QKD) allows for secure communication, but practical implementation faces challenges in the finite-size regime (where a finite number of signals $n$ are exchanged).

• Context: Continuous-Variable (CV) QKD protocols using Discrete Modulation (DM) (e.g., BPSK, QPSK) are attractive due to compatibility with existing telecom infrastructure. However, proving their security against general attacks is difficult because the Hilbert space is infinite-dimensional.
• The Gap: While asymptotic security (infinite $n$) is well-understood, finite-size security requires tight lower bounds on the smooth min-entropy to estimate the secret-key rate. Existing bounds, particularly those based on the Asymptotic Equipartition Property (AEP), often become loose or vanish for small block sizes ($n$), rendering them impractical for real-world, short-duration sessions.
• Goal: The authors aim to compute and compare different quantum conditional entropy bounds to derive tighter, more reliable secret-key rate estimates for CV-QKD with discrete modulation under collective Gaussian attacks (passive loss and thermal noise) in the finite-size regime.

2. Methodology

The authors analyze two specific protocols:

1. BPSK (Binary Phase-Shift Keying): Alice encodes bits into two coherent states; Bob uses homodyne detection.
2. QPSK (Quadrature Phase-Shift Keying): Alice encodes 2 bits into four coherent states; Bob uses heterodyne detection.

They consider two channel models:

• Pure-Loss Channel: Models attenuation in optical fibers. Eve collects the leaked field (passive attack).
• Thermal-Noise Channel: Models excess noise (e.g., imperfect detectors or environmental noise). Eve injects thermal photons (active Gaussian attack).

Key Technical Steps:

• Entropy Definitions: The paper utilizes four types of quantum conditional entropies to bound the smooth min-entropy ($\tilde{H}^{\epsilon}_{\min}$):
  1. Petz-Rényi entropy ($H^{\downarrow}_{\alpha}$)
  2. Optimized Petz-Rényi entropy ($H^{\uparrow}_{\alpha}$)
  3. Sandwiched Rényi entropy ($\tilde{H}^{\downarrow}_{\alpha}$)
  4. Optimized Sandwiched Rényi entropy ($\tilde{H}^{\uparrow}_{\alpha}$)
• Analytical Derivation: For the pure-loss scenario, the authors exploit the symmetry of the coherent states and the unitary group action to derive closed-form analytical expressions for these entropies. This avoids intensive numerical optimization for every parameter set.
• Numerical Estimation: For the thermal-noise scenario, the conditional states become mixed (non-pure), preventing closed-form solutions. The authors employ a dimension-reduction technique (truncating the infinite-dimensional Hilbert space to a finite Fock basis) combined with a penalty term ($\Delta(\omega)$) to ensure security bounds remain valid for the original infinite-dimensional system.
• Comparison Framework: They compare three distinct estimators for the secret-key rate ($r$):
  • $r_S$: Based directly on the optimized sandwiched Rényi entropy (Eq. 3).
  • $r_B$: Based on a continuity bound relating Rényi entropy to von Neumann entropy (Eq. 9).
  • $r_{AEP}$: Based on the traditional Asymptotic Equipartition Property (Eq. 4).

3. Key Contributions

1. Analytical Expressions for Pure-Loss: The paper provides the first analytical (or semi-analytical) expressions for Petz-Rényi and sandwiched Rényi conditional entropies specifically for BPSK and QPSK CV-QKD protocols under passive attacks.
2. Superior Finite-Size Bounds: The authors demonstrate that the bound derived from the optimized sandwiched Rényi entropy ($r_S$) is significantly tighter than the AEP-based bound ($r_{AEP}$) and the continuity bound ($r_B$) for small block sizes.
3. Thermal Noise Extension: They extend the analysis to thermal-noise channels, providing a method to calculate key rates using dimension reduction while accounting for the "weight" of the state outside the truncated subspace.
4. Optimization Insights: The study identifies optimal protocol parameters (coherent state amplitude $|\alpha|$ and Rényi parameter $\alpha$) that maximize the key rate. Notably, for very small $n$, the optimal $\alpha$ shifts away from 1, indicating a transition toward min-entropy dominance.

4. Key Results

• Performance in Pure-Loss (Fig. 3):
  • For large $n$ (asymptotic limit), all three estimators converge to the same value.
  • For small $n$ (finite-size), $r_S$ outperforms the others significantly.
  • Specifically, for BPSK with $\eta=0.9$, $r_{AEP}$ drops to zero when $n < 10^4$, whereas $r_S$ remains positive and non-vanishing even for $n \sim 500$ (yielding a rate of $\sim 0.09$).
  • The optimal Rényi parameter $\alpha$ increases as $n$ decreases, eventually approaching the min-entropy limit.
• Performance in Thermal Noise (Fig. 5, 6, 7):
  • The Rényi-based estimator ($r_{SD}$) is more robust against losses and noise compared to the AEP estimator.
  • In a thermal-noise channel with excess noise $\xi=0.01$, the AEP bound vanishes at $\sim 60$ km ($n=10^7$), while the Rényi bound sustains positive rates up to $\sim 140$ km.
  • Compared to previous works (e.g., Kanitschar et al. [13]), the authors' method yields slightly higher rates in the asymptotic regime (due to specific attack assumptions) but demonstrates superior endurance in the small block-size regime.
• Optimal Parameters:
  • BPSK: Optimal amplitude $|\alpha| \approx 0.93-0.99$ for pure loss.
  • QPSK: Requires higher amplitude ($|\alpha| \approx 1.58-1.67$) to achieve optimal rates compared to BPSK.

5. Significance

• Practicality for Short Sessions: The results are crucial for real-world QKD deployments where block sizes may be limited by channel coherence times or latency requirements. The ability to generate secure keys with $n < 10^4$ is a major advancement over AEP-based methods.
• Benchmarking Tool: The derived analytical expressions serve as "ball-park figures" and upper bounds for unknown channels, providing a benchmark for experimentalists.
• Theoretical Advancement: The work bridges the gap between discrete-variable (DV) QKD security proofs (which heavily utilize Rényi entropies) and CV-QKD, suggesting that sandwiched Rényi entropies are the optimal tool for finite-size CV security analysis.
• Future Direction: The paper paves the way for more complete security analyses that do not assume known channel parameters, potentially utilizing the dual representations of entropies provided in the appendices for device-independent or measurement-device-independent (MDI) scenarios.

In summary, this paper establishes that sandwiched Rényi entropies provide the most reliable and tightest lower bounds for secret-key rates in discrete-modulated CV-QKD, particularly in the challenging finite-size regime where traditional methods fail.